Including dispersion in density functional theory for adsorption on flat oxide surfaces, in metal-organic frameworks and in acidic zeolites

We examine the performance of nine commonly used methods for including dispersion interactions in density functional theory (DFT): three different parametrizations of damped 1/Rn terms (n = 6, 8, ...) added to the DFT energy (Grimme's D2 and D3 parameterizations as well as that of Tkatchenko and Scheffler), three different implementations of the many-body dispersion approach (MBD, MBD/HI and MBD/FI), the density-dependent energy correction, called dDsC, and two "first generation" van der Waals density functionals, revPBE-vdW and optB86b-vdW. As test set we use eight molecule-surface systems for which agreement has been reached between experiment and hybrid QM:QM calculations within chemical accuracy limits (±4.2 kJ mol-1). It includes adsorption of carbon monoxide and dioxide in the Mg2(2,5-dioxido-1,4-benzenedicarboxylate) metal-organic framework (Mg-MOF-74, CPO-27-Mg), adsorption of carbon monoxide as well as of monolayers of methane and ethane on the MgO(001) surface, as well as adsorption of methane, ethane and propane in H-chabazite (H-CHA). D2 with Ne parameters for Mg2+, D2(Ne), MBD/HI and MBD/FI perform best. With the PBE functional, the mean unsigned errors are 6.1, 5.6 and 5.4 kJ mol-1, respectively.